Google Health lets users turn off its AI Coach without canceling Premium: the switch that reveals AI-advisor fatigue

By Engadget · November 8, 2026.

Google Health now lets users turn off AI Coach, the Gemini-based assistant included in its Premium subscription, without canceling the plan. According to Engadget, the option arrived with an August 2026 update and can be enabled from the app's privacy controls.

According to Engadget's guide, the path is: profile picture, "Your data in Google Health," "Feature privacy controls," "Google Health Coach" and the off switch. Users can also go to Settings > Coaching, or switch the "Today" messages from "Standard" to "No Coach insights." A middle-ground option is to mute only the coaching notifications.

Turning it off has clear costs. Users lose personalized recommendations, summaries and the "Ask Coach" button. Meals can no longer be logged by photo, though they can still be entered manually. Existing data is not deleted. To turn it back on, users tap the "Ask Coach" button, which remains visible, and choose "Unlock Coach." They can't re-enable it from the same menu.

The rest of the Premium perks remain: personalized sleep summaries, advanced sleep-tracking algorithms, Sleep Score, Health Guardian (which monitors notable health changes) and cycle predictions, among others. According to the author, the Coach isn't available in the free version and doesn't offer medical advice.

The author admits that the Coach has been useful to her, for example for logging weights in strength training or adapting routines to a shoulder injury. She also acknowledges that it can get repetitive: every day it recommends things she already does. This is one person's experience, not usage data, and should be read that way.

**Our take.** This story is a practical guide, but it points to a broader thesis: the first problem with AI coaching isn't accuracy but saturation. An assistant that keeps repeating advice ends up ignored, and then it stops adding value. Google's decision to add an off switch shows that users want to decide how much AI enters their daily lives.

The design details matter too. Being able to turn off the feature without losing the rest of the service is good practice for user control. Having to turn it back on through a different path is minor friction, but it's the kind that keeps some users from coming back. And keeping the data after the feature is turned off is the sensible approach on privacy.

Broadly speaking, personal health assistants are at an early stage. Today they help with logging, motivation and adjusting routines, and they warn that they're no substitute for a doctor. It's low-risk territory, which makes it a good testing ground.

In the short term, we'll see plenty of friction: too many notifications, privacy concerns over sensitive data and subscriptions charging for features not everyone wants. In the long term, the promise is far greater. A companion that knows your history, picks up early warning signs and sensibly adjusts your habits could help prevent diseases before they appear. That future, one of continuous and accessible preventive medicine, will only become reality if users trust these tools and stay in control. Paradoxically, an off switch is part of that trust.

The bar for these products should be simple: useful when needed, quiet when not, and always possible to turn off.